If you run your python.exe under gdb you should be able to set a future breakpoint on your _PyEval_EvalMiniFrameEx function and debug from there.<br><br><div class="gmail_quote">On Wed, Jan 14, 2009 at 8:28 PM, <span dir="ltr"><<a href="mailto:skip@pobox.com">skip@pobox.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br>
I've recently been working on generating C functions on-the-fly which inline<br>
the C code necessary to implement the bytecode in a given Python function.<br>
For example, this bytecode:<br>
<br>
>>> dis.dis(f)<br>
2 0 LOAD_FAST 0 (a)<br>
3 LOAD_CONST 1 (1)<br>
6 BINARY_ADD<br>
7 RETURN_VALUE<br>
<br>
is transformed into this rather boring bit of C code:<br>
<br>
#include "Python.h"<br>
<br>
#include "code.h"<br>
#include "frameobject.h"<br>
#include "eval.h"<br>
#include "opcode.h"<br>
#include "structmember.h"<br>
<br>
#include "opcode_mini.h"<br>
<br>
PyObject *<br>
_PyEval_EvalMiniFrameEx(PyFrameObject *f, int throwflag)<br>
{<br>
<br>
static int jitting = 1;<br>
<br>
PyEval_EvalFrameEx_PROLOG1();<br>
co = f->f_code;<br>
PyEval_EvalFrameEx_PROLOG2();<br>
<br>
oparg = 0;<br>
LOAD_FAST_IMPL(oparg);<br>
oparg = 1;<br>
LOAD_CONST_IMPL(oparg);<br>
BINARY_ADD_IMPL();<br>
RETURN_VALUE_IMPL();<br>
<br>
PyEval_EvalFrameEx_EPILOG();<br>
}<br>
<br>
The PROLOG1, PROLOG2 and EPILOG macros are just chunks of code from<br>
PyEval_EvalFrameEx.<br>
<br>
I have the code compiling and linking, and dlopen and dlsym seem to work,<br>
returning apparently valid pointers, but when I try to call the function I<br>
get<br>
<br>
Program received signal EXC_BAD_ACCESS, Could not access memory.<br>
Reason: KERN_PROTECTION_FAILURE at address: 0x0000000c<br>
0x0058066d in _PyEval_EvalMiniFrameEx (f=0x230d30, throwflag=0) at MwDLSf.c:17<br>
<br>
Line 17 is the PROLOG1 macro. I presume it's probably barfed on the very<br>
first instruction. (This is all on an Intel Mac running Leopard BTW.)<br>
<br>
Here are the commands generated to compile and link the C code:<br>
<br>
gcc -fno-strict-aliasing -DNDEBUG -g -fwrapv -O3 -Wall \<br>
-Wstrict-prototypes -g -DPy_BUILD_CORE -DNDEBUG \<br>
-I/Users/skip/src/python/py3k-t/Include \<br>
-I/Users/skip/src/python/py3k-t -c dTd5cl.c \<br>
-o /tmp/MwDLSf.o<br>
gcc -L/opt/local/lib -bundle -undefined dynamic_lookup -g \<br>
/tmp/dTd5cl.o -L/Users/skip/src/python/py3k-t -lpython3.1 \<br>
-o /tmp/MwDLSf.so<br>
<br>
(It just uses the distutils compiler module to build .so files.) The .so<br>
file looks more-or-less ok:<br>
<br>
% otool -L /tmp/MwDLSf.so<br>
/tmp/MwDLSf.so:<br>
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 111.1.3)<br>
<br>
though nm doesn't show that any undefined _Py* symbols so I suspect I'm not<br>
linking it correctly. The Python executable was built without<br>
--enable-shared. I've tried building with that config flag, but that just<br>
gives me fits during debugging because it always wants to find libpython in<br>
the installation directory even if I'm running python.exe from the build<br>
directory. Installing is a little tedious because it relies on a properly<br>
functioning interpreter.<br>
<br>
dlopen is called very simply:<br>
<br>
handle = dlopen(shared, RTLD_NOW);<br>
<br>
I used RTLD_NOW because that's what sys.getdlopenflags() returns. I'm not<br>
calling dlclose for the time being.<br>
<br>
I'm not exactly sure where I should go from here. I'd be more than happy to<br>
open an item in the issue tracker. I was hoping to get something a bit<br>
closer to working before doing that though. The failure to properly load<br>
the compiled function makes it pretty much impossble to debug the generated<br>
code beyond what the compiler can tell me.<br>
<br>
Any suggestions?<br>
<br>
Skip<br>
_______________________________________________<br>
Python-Dev mailing list<br>
<a href="mailto:Python-Dev@python.org">Python-Dev@python.org</a><br>
<a href="http://mail.python.org/mailman/listinfo/python-dev" target="_blank">http://mail.python.org/mailman/listinfo/python-dev</a><br>
Unsubscribe: <a href="http://mail.python.org/mailman/options/python-dev/greg%40krypto.org" target="_blank">http://mail.python.org/mailman/options/python-dev/greg%40krypto.org</a><br>
</blockquote></div><br>